NVIDIA today released Nemotron 3.5 Super VL, an open 120-billion-parameter model, and two new edge models: Nemotron 3.5 Bolt 74B-A9B 24GB and Nemotron 3.5 Bolt E4B 4GB. Super VL delivers the fastest task completion and highest accuracy for specialized always-on agents, being 5x faster than competitors and roughly one-fifth the size of NVIDIA's largest reasoning model. The Bolt models are optimized for local agentic workloads, with Bolt E4B scheduled for release on Oct. 23.